Towards a Microbial Thermoelectric Cell

Raul Rodrguez-Barreiro1, Christian Abendroth1, Cristina Vilanova1, Andres Moya1,2, Manuel Porcar1,3*
1 Cavanilles Institute of Biodiversity and Evolutive Biology, Universitat de Valencia, Valencia, Spain, 2 Unidad Mixta de Investigacion en Genomica y Salud, Centro Superior
de Investigacion en Salud Publica (CSISP), Valencia, Spain, 3 Fundacio General de la Universitat de Valencia, Valencia, Spain

Abstract
Microbial growth is an exothermic process. Biotechnological industries produce large amounts of heat, usually considered
an undesirable by-product. In this work, we report the construction and characterization of the first microbial thermoelectric
cell (MTC), in which the metabolic heat produced by a thermally insulated microbial culture is partially converted into
electricity through a thermoelectric device optimized for low DT values. A temperature of 41uC and net electric voltage of
around 250600 mV was achieved with 1.7 L bakers yeast culture. This is the first time microbial metabolic energy has been
converted into electricity with an ad hoc thermoelectric device. These results might contribute towards developing a novel
strategy to harvest excess heat in the biotechnology industry, in processes such as ethanol fermentation, auto thermal
aerobic digestion (ATAD) or bioremediation, which could be coupled with MTCs in a single unit to produce electricity as a
valuable by-product of the primary biotechnological product. Additionally, we propose that small portable MTCs could be
conceived and inoculated with suitable thermophilic of hyperthermophilic starter cultures and used for powering small
electric devices.

Citation: Rodrguez-Barreiro R, Abendroth C, Vilanova C, Moya A, Porcar M (2013) Towards a Microbial Thermoelectric Cell. PLoS ONE 8(2): e56358. doi:10.1371/
journal.pone.0056358
Editor: Chenyu Du, University of Nottingham, United Kingdom
Received October 17, 2012; Accepted January 14, 2013; Published February 26, 2013
Copyright: 2013 Rodrguez-Barreiro et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Financial support was provided by grant Prometeo/2009/092 (Conselleria dEducacio, Generalitat Valenciana, Spain) to AM. The funders had no role in
study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have a patent describing the results they report on their article: The reference of the patent is P201200977 (application
number for the Spanish Office of Patents and Trademarks, OEPM). The authors of the patent are the same of the manuscript submitted to PLOS ONE: Porcar, M;
Rodrguez-Barreiro, R; Abendroth, C; Vilanova, C; and Moya, A. The complete title is Dispositivo termoelectrico microbiano y metodo asociado a dicho
dispositivo(Thermoelectric Microbial device and associated method). The authors have prepared the patent and the registration in collaboration with the
Research Transfer Office (OTRI) of the University of Valencia (contact person, Marta Garces: marta.garces@uv.es). The authors formally confirm that this patent
does not alter the authors adherence to all the PLOS ONE policies on sharing data and materials.
* E-mail: manuel.porcar@uv.es

Introduction anode in a simple device known as a Microbial Fuel Cell (MFC).

Both developed and fast growing developing countries exhibit
steadily growing energy demands. Taking into account the limited
nature of oil, coal and gas reservoirs, this could obviously lead to a
shortage of standard (fossil) fuels in the relatively near future. The
lack of sustainability of current fossil-centered energy strategies, as
well as the recent extremely serious accident at Fukushima Daiichi
power facility [1] have increased the concerns about the economic
and environmental consequences of relying on these energy
sources, leading to some dramatic shifts in energy policies, like in
Germany [2]. It is widely accepted that massive fossil fuel
consumption, which results in the production of nine billion metric
tons of atmospheric carbon per year [3], is at least partially
responsible for current global warming. Therefore, alternative
non-fossil non-nuclear technologies are seen as promising, albeit
not fully competitive. Among these, biomass-based energy has
been suggested as one of the most promising technologies for
renewable energy production [4,5]. Biomass from crops; urban,
industrial or agricultural wastes; green algae, cyanobacteria or
other microbial cultures, are renewable organic resources that are
suitable for energy production in the form of biofuels (mainly, but
not limited to, bioethanol and biodiesel), and electricity.
Besides lignocellulosic combustion-based power production, a
biological system allowing direct conversion of biomass into
electricity already exists: a broad range of organic substances can
be oxidized by electrogenic bacteria, which transfer electrons to an
At the cathode, other useful products can be generated, including
hydrogen, methane, and hydrogen peroxide [6,7,8]. The electric
yield of MFCs has increased dramatically in recent years, mainly
by increasing the ratio of the area of the electrodes/volume in the
reactor, with best yields reaching up to 27 W/m2. A moderate
MFC unit, of about 1 L, can produce enough electricity to power a
small propeller for more than one year [9]. However, MFCs seem
to work better at small scales, as scaling-up faces important
challenges [9].
Many bacterial species have been reported to display electro-
active properties, including members of common genera such as
Clostridium, Pseudomonas, Geobacter or Shewanella. A few eukaryotic
microorganisms have been assayed for power production in
MFCs. Bakers yeast Saccharomyces cerevisisae has proven able to
transfer electrons to an anode in two independent studies [10,11]
with moderate efficiency. In both reports, researchers found net
voltage values of about 0.33 V for 1 L reactors.
To date MFCs are still the only direct method to microbiolog-
ically convert biomass into electricity. Nonetheless, there is
possibly another non-fuel alternative. Since microbial growth is
an exothermic process, it produces heat, which is a by-product that
usually goes unnoticed in lab-scale cultures but which has a strong
impact on the design and performance of industrial-scale
microbial fermentations. Almost 90% of the heat produced in a
microbial fermentation is reported to be metabolic heat; and
almost all this heat is removed through forced heat exchange [12].

PLOS ONE | www.plosone.org 1 February 2013 | Volume 8 | Issue 2 | e56358


The thermoelectric or Peltier-Seebeck effect is the direct
conversion of electric voltage to temperature differences (Peltier
effect) and vice-versa (Seebeck effect). Theoretically, an electric
current would be produced by coupling an exothermic microbial
culture with an endothermic reaction or, alternatively, a heat
sink through a thermoelectric cell. If the thermal energy from
exothermic microbial cultures could be turned into electricity
efficiently, power-producing devices could be designed and
coupled to existing microbial reactors within a range of
applications (alcoholic fermentations, bioremediation, waste treat-
ment, autotrophic thermal aerobic digestion ATAD, etc.).
Here, we report the characterization of the first Microbial
Thermoelectric Cell, a bioreactor specifically designed for power
production through a completely different mechanism than that
operating in MFCs: the thermoelectric effect. Our results might
contribute to providing a new scenario for the future development
of microbial-based cellular electricity facilities, which might be
useful for local electric production and heat recycling in a wide
range of biological processes.
Materials and Methods
Construction of the MTC
In order to implement a thermoelectric-based power generator,
a reactor was designed able to i) sustain microbial growth; ii)
remain thermally isolated on most of its surface; and iii) efficiently
transfer heat through a relatively small area to a thermoelectric
device. One of us (M. Porcar) had previously designed an LCC
(Liquid Culture Calorimeter) for microbial growth, suitable for
fine recordings of internal temperature changes through a
thermocouple [13]. Based on the LCC, we conceived a Microbial
Thermoelectric Cell (MTC hereon) to produce power from a
microbial culture by the Peltier-Seebeck effect. Figure 1 shows the
structure of the MTC. The core of the reactor is a 1.9 L glass
container from a commercial vacuum flask. The flask was placed
inside an expanded polystyrene (EPS) box and the gap filled with a
polyurethane foam spray (Silicex Fischer, Fisher Iberica, Tarra-
gona, Spain). The box was then inserted into a second EPS
isolation box. The upper part of the MTC was drilled and a
cupper bar (20 mm in diameter) inserted through the hole. The
upper part of the cupper bar was adapted in order to allow a TE-
Power Probe thermal harvester (MicroPelt, Germany) to be
screwed through a 1/40 Whitworth thread (DIN 2999, JIS B0203,
ISO 7/1). TE-Power Probe is a prototype of an integrated
proximity thermoharvester designed to replace primary batteries
in wireless systems operating in duty cycle mode. The key element
of the TE-Power Probe is the MPG-D751 thermogenerator, which
produces electricity from a rather low gradient of temperature.
The TE-Power Probe is originally designed to attach to a heat
source in the shape of piping that carries a hot fluid, and heat is
dissipated through an aluminum heat sink, with the resulting
temperature gradient allowing power production by the MPG-
D751 thermogenerator. In our experiments, temperature changes
in different parts of the Probe were measured by PT-100 sensors.
Since the TE-Power Probe is specifically designed to operate using
natural convection to ambient air, we mounted it horizontally, as
suggested by the manufacturer.
The two EPS isolation layers of the MTC were shaped so the
round bottom of the vacuum flask would fit. The flask bottom was
placed conveniently close (20 mm) to the bottom of the MTC in
order to allow stirring by a magnetic stirrer. When recordings were
to be taken, the MTC was first filled with 1.8 L of medium; a small
magnet was added; the MTC was placed inside a standard
laboratory magnetic stirrer set at low speed (600 rpm); the
PLOS ONE | www.plosone.org
Towards a Microbial Thermoelectric Cell
inoculum was then added, and the cupper bar with the screwed
TE-Power Probe finally set in place. This configuration was
modified for characterization purposes in some experiments, as
described in sections 2.4 and 2.5.
Yeast Strains, Media and Culture Conditions
The following six diploid S. cerevisiae strains, from the wine
industry or genetic modifications thereof, were used: EC118,
L2056, 3aS2D, T73, D170, and TTRX2. All the strains were
kindly provided by Prof. Emilia Matallana (IATA, Valencia,
Spain). In order to assess their exothermic abilities, independent
cultures were set in filter-sterilized YPD (20 g/L peptone, 10 g/L
yeast extract, with 18% sugar), and the internal temperature of the
cultures (grown overnight in non-isolated Erlenmeyer flasks) was
continuously measured. Thermotolerance was assessed by growing
the strains at 30, 37 and 41uC. After an overnight incubation
under low stirring, the OD600 of the six overnight cultures was
measured.
For standard experiments after strain selection, the filter-
sterilized 18% sucrose YPD was inoculated with 1:50 of an
overnight yeast pre-culture grown at 30uC, and subjected to low
stirring for 120 h.
Data Acquisition, Monitoring and Recording
The MTC was connected to a PC in order to record internal
and external temperatures and the output current provided by the
heat harvester TE-Power Probe (Fig. S1).
The internal temperature of the MTC was measured by a thin
T-type thermocouple inserted into the microbial culture and
connected to a PC through a data logger, as previously described
[13]. Another thermocouple recording room temperature was also
set in place. The thermocouples were connected to an acquisition
card inserted on the data logger, which was connected via a GPIB
cable to a PC with an acquisition software that one of us (R.
Rodrguez-Barreiro) conceived specifically for this work (Fig. S1).
The TE-Power Probe output was also connected to the PC, which
yielded two additional temperature recordings by using two Pt-100
sensors (that of the cold and hot sides of the thermogenerator
device) and the output voltage. The connections between the
thermocouples and the data logger were performed on an ice-
water mixture to take into account the unwanted background
electric voltage, due to the junction of dissimilar metals in the
thermocouple-data logger connection. Finally, a thermocouple
was inserted inside the box containing the ice-water mixture in
order to verify that the temperature of the datalogger-thermocou-
ple connections was kept at 0uC.
Temperature records (and, when TE-Power Probe was
connected, electric power) were taken every 6 minutes throughout
the experiment.
Identification Assay to Estimate Broth Heat Capacity and
Global Thermal Resistance of MTC
In order to estimate the heat capacity of the broth (m?Cp) and the
global thermal resistance (Rg), the MTC (without TE-Power Probe)
was set up under the following conditions: first, an electrical
resistance was placed inside the MTC to generate a controlled
heat flow, as consequence of the Joule effect induced by an
external voltage input through the resistance. Second, the MTC
was loaded with room-temperature sterile broth with 1 g/L
nipagine supplementation to avoid contamination by yeasts. Broth
was subjected to continuous stirring and room temperature was
kept constant. Throughout the experiment, broth and room
temperatures and the power generated in the resistance were
2 February 2013 | Volume 8 | Issue 2 | e56358
 Towards a Microbial Thermoelectric Cell

Figure 1. Schematic drawing of the Microbial Thermoelectric Cell (Auto-CAD). All dimensions are given in mm.
doi:10.1371/journal.pone.0056358.g001

continuously measured. To ensure the initial steady-state condi-

tions (broth temperature equal to room temperature), the system X X X
Heat accumulated ~ Heat generated{ Heat lost 1
was kept in the off mode for approximately 20 h before applying
the input voltage.
Heat accumulated is a consequence of the variation in broth
Theory temperature. Since there is no forced cooling of the system, heat
The equation for the heat flow balance corresponding to the flow losses are due only to heat transfer from the culture to the
MTC we describe in this work can be stated as follows:

PLOS ONE | www.plosone.org 3 February 2013 | Volume 8 | Issue 2 | e56358

 Towards a Microbial Thermoelectric Cell

environment through both the MTC surface and the TE-Power
Probe thermogenerator. For calibration purposes, we first set the
MTC to generate a heat flow from an electric resistance placed
inside the vacuum flask through the Joule effect. In standard
experiments, heat flow was obtained from the metabolic heat as a
consequence of microbial growth.
Therefore, the heat flow balance equation can be written for the
MTC as follows (a definition of all the symbols used throughout
the MTC modelling is available in Table 1):
the environment; and QTh is the heat flow loss through the cupper
bar connected to the TE-Power Probe.
Accumulation Terms
Heat accumulation (Qacc) in a particular body is determined by
the variation in its temperature (dTi/dt) and by its heat capacity
(mi?Cpi). In the MTC, heat can be accumulated in the broth
(subscript b), the vacuum flask (v) and the insulation walls
(w), as follows:

 
_ p {Q
_ acc ~ PJ zQ _ env {Q
_ Th _ acc ~ mb :Cp : dTb zmv :Cp : d T v zmw :Cp : d T w
Q 3
Q 2 b dt v
dt w
dt
Where Qacc is the heat flow accumulated in the broth; PJ is the heat
flow due to the Joule effect; Qp is the heat flow due to microbial The MTC is a very simple system with a single sensor to
metabolism; Qenv is the heat flow loss through the MTC surface to measure the temperature of the broth. Therefore, the equation can

Table 1. Nomenclature used in MTC modelling.

Symbol Description (units)

a Seebeck coefficient (V/K)

I Electrical current (A)
m? Cp Whole system heat capacity (J/K)
mb?Cpb Broth heat capacity (J/K)
mv? Cpv Vacuum flask heat capacity (J/K)
mw? Cpw Insulation walls heat capacity (J/K)
PJ Electrical input power due to the Joule effect (W)
Pe Electrical power generated (W)
Qacc Accumulated heat flow (W)
QC Net heat flow released through the cold side of the thermogenerator (W)
Qenv Heat flow released to the environment (W)
QH Net heat flow absorbed through the hot side of the thermogenerator (W)
Qj Heat flow due to the Joule effect inside of the thermogenerator (W)
Qp Heat produced by microbial metabolism (W)
QsC Heat flow produced in the cold side of the thermogenerator due to the Seebeck effect (W)
QsH Heat flow produced in the hot side of the thermogenerator due to the Seebeck effect (W)
Qt Heat flow loss due to the natural thermal conduction established between both sides of the thermogenerator (W)
QTh Heat flow absorbed from the broth through the cupper bar (W)
R Electrical resistance (V)
RCu Thermal resistance of the cupper bar (K/W)
Rg Global thermal resistance of the MTC (K/W)
Ri Internal electrical resistance of the thermogenerator (V)
RLoad Electrical resistance connected between the terminals of the thermogenerator (V)
RSk Thermal resistance of the heat sink (K/W)
Rth Thermal resistance of the thermogenerator (K/W)
Tb Broth temperature (K)
TC Temperature of the cold side of the thermogenerator (K)
Tenv Room temperature (K)
TH Temperature of the hot side of the thermogenerator (K)
DTth Difference of temperature between the hot and the cold sides of the thermogenerator (K)
Tv Vacuum flask temperature (K)
Tw Insulation walls temperature (K)
Vext Input voltage (V)
Vo Voltage output in the terminals of the thermogenerator (V)

doi:10.1371/journal.pone.0056358.t001

PLOS ONE | www.plosone.org 4 February 2013 | Volume 8 | Issue 2 | e56358


be simplified:
_ acc & m:Cp : dTb
Q 4
dt
Where m?Cp represents the whole system heat capacity, deduced
from the variation in culture temperature. This parameter can
easily be determined under a simplified experimental configura-
tion (described in 2.4) using the model equations described below
(section 3.2).
Loss Terms (I): Heat Flow Loss to the Environment
Energy losses through the MTC walls can be due to the natural
heat flow (Qenv) from the warm internal broth to the relatively cool
environment. Since insulation materials in the MTC display low
emissivity values, radiation losses can be neglected and Qenv can be
expressed as follows:
_ env ~ 1 :Tb {Tenv
Q 5
Rg
Where Rg represents the global thermal resistance of the MTC
and Tb and Tenv are the temperatures of the broth and the
environment, respectively. This thermal resistance can be exper-
imentally determined under the same conditions described for
m?Cp (see 2.4 and 3.2).
Loss Terms (II): Heat Flow Loss Through the TE-Power
Probe
The global heat flow through the cupper bar (QTh) is the same
than the heat flow absorbed by the hot side of the thermogenerator
cell (QH) and is composed of: (i) a spontaneous flow due to the
difference in temperature between the hot and cold sides of the
thermogenerator cell, expressed as (TH-TC)/Rth; (ii) an induced
heat flow as a consequence of the conversion of heat to electric
power through the Seebeck effect. Then, the heat flow loss
through the thermogenerator can be stated as follows [14]:
    the gain (Rg) and the time constant (m?Cp?Rg) were estimated from
Q _ H ~ TH {TC z a:TH :I{ 1 :I 2 :Ri
_ Th ~ Q 6 the experimental values of Tb and PJ.
Rth 2
Where a?TH?I corresponds to heat absorbed by the thermo-
generator due to the Seebeck effect, while the term 1/2?I2?Ri
corresponds to the heat produced as a consequence of the Joule
effect, associated to the circulation of the electric current produced
through the internal resistance of the thermogenerator. TH and TC
represent the temperature of the hot and cold sides of the cell,
whereas Rth and Ri correspond to its internal thermal and electrical
resistance, respectively. a is the Seebeck coefficient of the
thermogenerator and I is the electrical current obtained from
the TE-Power Probe.
Generation Terms (I): Heat Flow Due to the Joule Effect
When an electrical resistance was placed inside the vacuum
flask, a heat flow (PJ) was obtained as a consequence of applying
an external voltage according to the Joule effect:
PLOS ONE | www.plosone.org
Towards a Microbial Thermoelectric Cell
2
Vext
PJ ~ 7
R
Where Vext is the input voltage and R is the electrical resistance.
Generation Terms (II): Heat Flow Due to Yeast Growth
When the electrical resistance was replaced by a yeast culture,
the heat flow was generated as a consequence of the exothermic
properties of yeast metabolism. This heat flow, represented as Qp,
was estimated for the different experimental configurations as
described below (section 3.3).
Taking all the equations described above together, the general
energy balance (Eq. 2) can be written as:
m:Cp :
dTb _ p { Tb {Tenv {Q
~ PJ zQ _ Th 8
dt Rg
Model Equations for the Estimation of m?Cp and Rg
In order to calculate the global heat capacity and the global
thermal resistance of the MTC (m?Cp and Rg, respectively), a
simplified experimental set up was used, as explained in section
2.4. Briefly, heat flow was induced in the sterile broth by applying
a constant input power through a resistance according to the Joule
effect. In this experiment, room temperature was kept constant
and the TE-Power Probe was not mounted on the MTC.
Therefore, QTh and Qp terms (corresponding to the TE-Power
Probe and the yeast, respectively) from Eq. 8 are null, so it can be
written as the following first-order EDO:
dTb t Tb t{Tenv
m: C p : ~ PJ { 9
dt Rg
A first-order EDO is mathematically characterized by its gain
and its time constant, which can be estimated manually or with a
standard mathematical software from experimental data. In Eq. 9,
Estimation of Heat Yield Due to Yeast Metabolism and
Calculation of the Electrical Power Generated
Heat yield due to yeast metabolism (Qp) was estimated from Eq.
8, where the term PJ is null since no electrical resistance was set up
inside the flask:
_ p t ~ m:Cp : dTb t z Tb t{Tenv t zQ
Q _ Th t 10
dt Rg
In the assays where the TE-Power Probe was not included, the
term QTh (the broth heat lost through the cupper bar) is null, so Qp
was calculated from the experimental data of Tb and Tenv using the
estimations of m?Cp and Rg previously obtained.
When the TE-Power Probe was included, the metabolic heat
yield was calculated from Eq. 10, along with the model equations
for TE-Power Probe in order to estimate QTh (a detailed
description of these equations and a schematic representation of
associated heat flows is available in Appendix S2 and Fig. S2,
5 February 2013 | Volume 8 | Issue 2 | e56358

Figure 2. Time course of broth and room temperatures during the identification assay of broth heat capacity and global thermal
resistance of the MTC. The experiment was carried out under the conditions described in section 2.4. Recordings of room temperature (blue),
broth temperature (red) and input power (dashed line) were taken every 6 min.
doi:10.1371/journal.pone.0056358.g002
respectively). These model equations are dependent on the
electrical configuration used in the thermogenerator during the
assays. When no load resistance was connected to the terminals of
the thermogenerator (no electrical power was taken out), the
following equation for TE-Probe was used (for a detailed version of
this open-circuit model, see Appendix S1):
_ Th t ~ DTth t
Q 11
RTh
DTth represents the difference of temperature between the hot
and the cold side of the thermogenerator, whereas Rth is the
thermal resistance of the thermogenerator.
Voltage output (Vo) of the terminals of the TE-Power Probe,
which under this configuration is equal to the voltage generated in
the Peltier cell, can be expressed as:
Vo t ~ a:DTth t 12
Being a the Seebeck coefficient.
Otherwise, when a load resistance was fitted to achieve the
maximum power from the thermogenerator, Eq.11 was replaced
by Eq.13 (deduced in the maximum-power model of Appendix
S2):
 
_ a2 :DTth t : TH t DTth t Ri
QTh t ~ { z 2: 13
Ri 2 8 a Rth
Where Ri and Rth are the internal electrical and thermal
resistance, respectively.
PLOS ONE | www.plosone.org
Towards a Microbial Thermoelectric Cell
Under this configuration, voltage output (Vo) of the terminals of
TE-Power Probe can be expressed as:
a:DTth t
VO t ~ 14
2
and the maximal power generated can be calculated as follows:
VO2 t
P e t ~ 15
Ri
Results
Estimation of Broth Heat Capacity and Global Thermal
Resistance of MTC
In order to characterize the thermal evolution of the MTC prior
to the experiments with yeast cultures, an identification assay for
m?Cp and Rg was set up as described in 2.4. The time course of
broth and room temperature during the experiment is shown in
Figure 2. From a steady-state, in which room and broth
temperature were the same (25.5uC), a constant input power of
1 W was supplied, and the broth reached a final temperature of
47.5uC. The system gain (meaning the temperature increase
divided by the input power) was 22 K/W, and the time constant
(the time by which 63% of the temperature increase is reached)
was 43.5 h. According to the model equations (see 3.2), the gain
represents the global thermal resistance of the MTC, and the
broth heat capacity can be obtained by dividing the time constant
by the gain. Thus, our estimated values for Rg and m?Cp are 22 K/
W and 7118 J/K, respectively.
When the mathematical software was used to estimate Rg and
m?Cp from the same experimental data, similar values were
6 February 2013 | Volume 8 | Issue 2 | e56358

Figure 3. Typical performance of the MTC without TE-Power Probe. Experimental values of broth and room temperature (red and blue lines,
respectively) are shown.
doi:10.1371/journal.pone.0056358.g003
obtained (Rg = 22 K/W and m?Cp = 7100 J/K) with a confidence
level of 98.7%.
Strain Selection and MTC Performance
All yeast strains exhibited similar performance in terms of
exothermic potential and resistance to high temperatures, with
strain D170 displaying slightly higher thermoresistance (data not
shown). This strain was thus selected for further studies. When
yeast strain D170 was inoculated into a pre-warmed 18% sucrose
YPD medium and grown in the MTC without the cupper bar and
the TE-Power Probe set in place, the internal temperature
dropped slowly (about 1uC), stabilized and finally started to rise
after 67 h. The temperature peaked after approximately 24 h
and reached up to 41uC. Figure 3 shows a typical experiment in
which the maximum temperature is around twelve degrees higher
than the initial temperature of the culture. After the peak, the yeast
culture temperatures started dropping and reached the initial
temperature after about 7090 h. Despite the abrupt changes
(22uC27uC) in room temperature as a consequence of switching
the air conditioning on and off, the change in the internal
temperature of the yeast culture was only mildly affected.
Estimation of Heat Yield Due to Yeast Growth
The heat yield due to yeast growth (Qp) was estimated as
described in section 3.3 for each experimental set up (Fig. 4). In all
the experiments, the estimated evolution of Qp peaked before broth
temperature reached its maximum due to the high inertia of the
broth (m?Cp). In the assay carried out without TE-Power Probe
(Fig. 4A), Qp reached its maximum (1.96 W) after 20 h and
remained above 0.2 W for 40 h. In an open-circuit configuration,
maximum Qp (almost 1.4 W) peaked after 10 h, reaching values
above 0.2 W over 50 h (Fig. 4B). Maximum Qp was obtained
earlier in this case because a more concentrated inoculum was
used, indicating that, as expected, there is a dependence between
initial yeast concentration and time until Qp maximum. Finally,
PLOS ONE | www.plosone.org
Towards a Microbial Thermoelectric Cell
under a load-resistance configuration, Qp peaked (with a value of
almost 1.5 W) after 20 h (as in the experiment without TE-Power
Probe, in which the same initial yeast concentration was used),
producing more than 0.2 W for 50 h (Fig. 4C). Our data show that
when the TE-Power Probe was inserted, lower Qp values were
estimated from experimental data. In accordance, total energy
generated by yeast metabolism, calculated as the area below the
curve of Qp, was higher in the experiment carried out without the
TE-Power Probe (194,7 kJ) in comparison with those configura-
tions in which it was included (144,4 and 145,4 kJ for the open-
circuit and the load-resistance set up, respectively). This might be
due to the effect of the cupper bar on effective broth stirring,
which might be lower and therefore affect yeast growth.
Electricity Production with the MTC
Under the MTC insulation conditions assayed, the metabolic
heat produced by strain D170 was partially transformed into
electricity through the TE-Power Probe thermal harvester. When
the TE-Power Probe was mounted in the yeast-culturing MTC
under open circuit conditions, the internal temperature of the
culture increased up to about 35uC and remained higher than
32uC for about 54 h (Fig. 5A). Under these conditions, electric
voltage yielded around 250 mV (net value) for a two-day period,
with significant, lower room temperature-associated peaks of about
350600 net mV (Fig. 5A). The same experiment was performed
under load resistance conditions (330 V, the same as that for the
MPG-D751 thermogenerator) and produced an internal temper-
ature peak of about 32uC, with the culture being hotter than room
temperature (which was constant in this experiment) for a period
of 110 h (Fig. 5B). Under these conditions, a maximum of 290 mV
were obtained on the electrical load resistance, corresponding to
around 580 mV generated in the Peltier cell (Eq. 15). The
maximum power obtained, corresponding to the maximum DT
values, reached around 255 mW (net value).
7 February 2013 | Volume 8 | Issue 2 | e56358
 Towards a Microbial Thermoelectric Cell

PLOS ONE | www.plosone.org 8 February 2013 | Volume 8 | Issue 2 | e56358

 Towards a Microbial Thermoelectric Cell

Figure 4. Time course of broth and room temperatures and heat yield due to yeast growth for different MTC configurations:
without TE-Power Probe (A), open-circuit (B) and load-resistance (C). Experimental values of broth and room temperature (red and blue
lines, respectively) were recorded every 6 min. Heat yield (dashed line) was estimated for each configuration as described in section 3.3.
doi:10.1371/journal.pone.0056358.g004

The energy conversion yield was calculated for this latter
experiment as the total electrical power generated (33.1 J) divided
by the total heat energy produced by the yeasts (147.44 kJ, as
calculated from the estimated heat yield represented in Fig. 4C).
The resulting value, 0.022%, was low, as expected from the poor
efficiency of heat-harvesting devices such as the TE-Power Probe.
Notwithstanding, it allowed the production of significant amounts
of electrical power from relatively moderate values of DT.
Discussion
The results presented here clearly indicate that the exothermic
nature of microbial growth can be exploited when transformed
into significant electric voltage. We have designed and constructed
the first Microbial Thermoelectric Cell, which consists of a simple,
thermally insulated reactor, with a small heat exchange surface on
which a thermoelectric prototype thermal harvester, equipped
with a MPG-D751 thermogenerator device (TE-Power Probe) is
mounted. The chosen thermogenerator is optimum for relatively
high efficiencies in electric production at low DT values, such as
those existing between an insulated yeast culture (41uC, under our
conditions) and room temperatures. With a medium size MTC
(smaller than two liters), we typically obtained 150600 mV.
These values are similar to those obtained with yeast-based MFCs
for which net voltage values of about 0.33 V for 1 L reactors have
been reported [10,11]. It is noteworthy that MFCs and MTCs
work on a totally different basis albeit theoretically compatible
as MFCs produce electricity from direct microbial-mediated
electron transfer from organic matter oxidation to an anode;
whereas the MTC partially transforms metabolic thermal energy
into electricity by the Seebeck effect. As it is also the case for
MFCs, MTCs could be combined with other microbial processes.
Bakers yeast S. cerevisiae was used for our MTC due to its well-
known exothermic growth under a range of different conditions.
Indeed, any other microbial culture resulting in important heat
production, such as ethanolic fermentation (beer, bread, wine,
biofuels), auto thermal aerobic digestion (ATAD) or hydrocarbon-
polluted soil bioremediation and bioaugmentation, could be
coupled with MTCs into a single unit, with electricity production

Figure 5. Electricity production by MTC under open-circuit (A) and load-resistance (B) configurations. The experimental temperature
values of broth (red), room (blue), and thermogenerator hot and cold sides (red and blue dashed lines, respectively) are shown along with the
evolution of voltage and power output (black continuous and dashed lines, respectively).
doi:10.1371/journal.pone.0056358.g005

PLOS ONE | www.plosone.org 9 February 2013 | Volume 8 | Issue 2 | e56358


as a valuable sub-product of the main biotechnological purpose. In
fact, metabolic heat is often seen in industry as an undesirable sub-
product of large-scale microbial fermentations, and cooling
facilities are often needed in order to maintain an optimum broth
temperature [12,15]. The conversion, albeit partial, of this heat
into electricity would both help to control internal temperatures in
biotechnological processes and contribute to energy savings by
cogeneration. Interestingly, our results suggest that heat produc-
tion through metabolic growth and heat flow through a
thermogenerator can be tuned in such a way that no energy is
needed to heat the broth up for microbial growth nor to cool it
down in order to avoid excessive temperatures, known to abruptly
stop the fermentation process.
It seems reasonable to predict that, in addition to yeast, other
cultures might be suitable as heat producers in an MTC. For
example, naturally-occurring thermophilic and hyperthermophilic
bacteria, such as Bacillus coagulans, Bacillus licheniformis or many
Geobacillus spp. strains, many of which can be isolated from extreme
environments such as deep oil wells and the optimal growth of
which is 5060uC. Additionally, these bacteria are reported as able
to heat their own culture up to 5055uC [16]. The perfect
candidate for MTC should meet the following criteria: (i)
thermoresistant; (ii) strong exothermic ability; (iii) rapid and easy
growth; and (iv) an ability to grow and degrade high concentra-
tions of carbon sources.
In the MTC we designed, the cold side of the system was an
aluminum heatsink. In order to optimize electricity yield by
increasing DT, a biological cooling system could theoretically be
implemented, rather than simple convection-driven heat loss. In
fact, methanogenic archaea have been reported to exhibit
endothermic growth [7]. Although it is uncertain whether
endothermia is a result of particular growth or of heat loss due
to gas evaporation from the culture, the fact is that these
microorganisms could be combined with those producing heat
through a thermoelectric element in order to increase electricity
production. These archaea have optimal growth at temperatures
of around 37uC, and this implies that the whole system should be
finely tuned in order to regulate heat transfer across the
thermoelectric element, allowing optimal microbial growth while
maintaining as high a DT as possible.
The surface:volume ratio of microbial fermentors is a critical
factor affecting heat loss to the environment and thus internal
temperature of the culture. Although standard lab-scale microbial
cultures produce heat, most of it is lost to environment due to high
surface to volume ratio, resulting in the absence of any noticeable
increase in internal temperature. However, large, production-scale
bioreactors have been characterized thermodynamically and
proved to work nearly adiabatically due to much lower surface
to volume ratio compared to laboratory-scale non-insulated
bioreactors [12]. The results presented here, together with
previous reports on medium-scale liquid culture calorimeters
[13], demonstrate that relatively small liquid cultures can also
work almost adiabatically, provided proper insulation is provided
and significant autothermal growth can be achieved. This implies
that small portable MTCs for electricity production could be
envisaged, since most of the metabolic heat from microbial growth
can be stored inside the MTC. These small thermoelectric cells
could theoretically be used to power small electric devices.
However, in order for MTCs to display higher electric yields,
optimization of the thermoelectric elements should take place.
Indeed, only 0.58% of the total heat flow is usually transformed
into electricity through the thermoelectric plates. Interestingly,
only 12% of the maximum theoretical efficiency is achieved in the
best thermoelectric devices today [17], so there is still room for
PLOS ONE | www.plosone.org
Towards a Microbial Thermoelectric Cell
significant improvement in the optimization of this technology.
There has been a dramatic increase in research into high efficiency
thermoelectric devices in recent decades, with reports of significant
improvements in ZT values, design optimization, and develop-
ment of alternative materials. As proposed by [17], TE solid-state
heat engines could well play a crucial role in addressing some of
the sustainability issues we face today.
Other heat harvesting methods, such as absorption heat
transformers or organic Rankine cycle, have been reported
previously [18,19]. However, these systems are space-consuming
and involve mobile parts that require continuous maintenance. In
contrast with these, solid-state thermoelectric systems are small,
require almost no maintenance, and display high adaptability to a
range of industrial designs [17].
In conclusion, this is the first report of microbial metabolic
energy being converted into electricity with an ad hoc thermoelec-
tric device, i.e., the Microbial Thermoelectric Cell. Our results
show that even small volumes of broth are able to exhibit
significant autothermal performance and produce electricity when
properly insulated and set in such a way that heat exchange is
minimized over the whole surface, except the small area on which
a (prototype) thermal harvester is mounted. Although the electric
power we obtained was rather low, this work may contribute
towards a novel strategy to harvest excess heat produced by the
biotechnology industry, particularly if ongoing research into
thermoelectric materials and design finally yields high efficiency
thermoelectric devices.
Supporting Information
Figure S1 Schematic drawing of MTC data-recording
system. Dashed lines represent thermocouple connections
measuring the temperature of the broth (Tb), the temperature of
the hot and cold sides of the thermogenerator (TH and TC,
respectively), and the room temperature (Tenv); whereas continuous
lines represent voltage measurements corresponding to the
thermogenerator (Vth) and the electrical resistance (Vr).
(TIF)
Figure S2 Schematic drawing of heat flows and resis-
tances within the thermogenerator cell. Symbols used are
in accordance with the nomenclature summarized in Table 1.
(TIF)
Appendix S1 Thermogenerator cell (MPG-D751) general
equations.
(DOCX)
Appendix S2 TE-Power Probe model description.
(DOCX)
Acknowledgments
We are very grateful to Emilia Matallana, for kindly supplying yeast strains,
Julian Heredero, for his fine work manufacturing the copper bar, to Ruslan
Klymenko for assistance with Figure 1 and to Fabiola Barraclough for
correction of the English text.
The technology described in this work has been found by us to hold not
only for scientific publication, but also for patenting (Application number
P201200977 at Spanish Office of Patents and Trademarks, OEPM).
Author Contributions
Conceived and designed the experiments: RRB CA CV MP. Performed
the experiments: RRB CA CV MP. Analyzed the data: RRB CA CV AM
MP. Contributed reagents/materials/analysis tools: AM. Wrote the paper:
RRB CV AM MP.
10 February 2013 | Volume 8 | Issue 2 | e56358

References
1. Dauer LT, Zanzonico P, Tuttle RM, Quinn DM, Strauss HW (2011) The
Japanese tsunami and resulting nuclear emergency at the Fukushima Daiichi
power facility: technical, radiologic, and response perspectives. J Nucl Med 52
(9): 14231432.
2. Gross M (2011) Energy U-turn in Germany. Curr Biol 21 (10): 379381.
3. Lehmann J (2007) A handful of carbon. Nature 447: 143144.
4. Kim D, Chang IS (2009) Electricity generation from synthesis gas by microbial
processes: CO fermentation to microbial fuel cell technology. Bioresour Technol
100: 45274530.
5. Song C (2002) Fuel processing for low-temperature and high-temperature fuel
cells: Challenges, and opportunities for sustainable development in the 21st
century. Catal Today 77 (1): 1749.
6. Cheng S, Xing D, Call DF, Logan BE (2009) Direct biological conversion of
electrons into methane by electromethanogenesis. Environ Sci Technol 43 (10):
39533958.
7. Liu H, Grot S, Logan BE (2005) Electrochemically assisted microbial production
of hydrogen from acetate. Environ Sci Technol 39 (11): 43174320.
8. Rozendal RA, Hamelers HVM, Euverink GJW, Metz SJ, Buisman CJN (2006)
Principle and perspectives of hydrogen production through biocatalyzed
electrolysis. Int J Hydrogen Energy 31 (12): 16321640.
9. Logan BE (2010) Scaling up microbial fuel cells and other bioelectrochemical
systems. Appl Microbiol Biotechnol. 85(6): 166571.
10. Gunawardena A, Fernando S, To F (2008) Performance of a yeast-mediated
biological fuel cell. Int J Mol Sci 9(10): 18931907.
PLOS ONE | www.plosone.org
Towards a Microbial Thermoelectric Cell
11. Ducommon R, Favre MF, Carrard D, Fischer F (2010) Outward electron
transfer by Saccharomyces cerevisiae monitored with a bi-cathodic microbial fuel
cell-type activity sensor. Yeast 27 (3): 139148.
12. Turker M (2004) Development of biocalorimetry as a technique for process
monitoring and control in technical scale fermentations. Thermochim Acta 419
(1): 7381.
13. Delas J, Notari M, Fores J, Pechuan J, Porcar M, et al. (2009) Yeast cultures with
UCP1 uncoupling activity as a heating device. N Biotechnol 26 (6): 300306.
14. Lineykin S, Ben-Yaakov S (2007) Modeling and analysis of thermoelectric
modules. IEEE Trans Ind Appl 43 (2): 505512.
15. von Stockar U, van der Wielen LAM (1997) Thermodynamics in biochemical
engineering. J Biotechnol 59 (1): 2537.
16. Ungwuanyi JO, Harvey LM, McNeil B (2008) Diversity of thermophilic
populations during thermophilic aerobic digestion of potato peel slurry. J Appl
Microbiol. 104 (1): 7990.
17. Bell LE (2008) Cooling, heating, generating power, and recovering waste heat
with thermoelectric systems. Science 321 (5895): 14571461.
18. Larjola J (1995) Electricity from industrial waste heat using high-speed organic
Rankine cycle (ORC). Int J Prod Econ 41 (1): 227235.
19. Saidur R, Rezaei M, Muzammil WK, Hassan MH, Paria S, et al. (2012)
Technologies to recover exhaust heat from internal combustion engines. Renew
Sust Energ Rev 16: 56495659.
11 February 2013 | Volume 8 | Issue 2 | e56358
Copyright of PLoS ONE is the property of Public Library of Science and its content may not be copied or
emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.
However, users may print, download, or email articles for individual use.